NuScale Power is worth $4.2 billion on the stock market, but the small modular reactor company generated only $10.7 million in revenue over the past 12 months. The massive gap reveals a fundamental bet: that AI data centers hungry for around-the-clock power will finally make nuclear energy economically viable, after the last U.S. nuclear expansion ended in near-total failure.

What Happened to the Last Nuclear Boom?

Between 2008 and 2009, the United States launched what was called a nuclear renaissance. Utilities filed combined license applications with the Nuclear Regulatory Commission (NRC) for 26 new reactors across 17 sites. The promise was clear: nuclear power would surge, and America would lead a global energy transformation.

The reality was brutal. By 2024, only two of those 26 reactors were ever completed. Georgia's Vogtle Units 3 and 4 were originally estimated to cost $14 billion and enter service in 2016 and 2017. Instead, they cost more than $30 billion and didn't begin commercial operation until July 2023 and spring 2024, roughly seven years late. South Carolina's V.C. Summer expansion started construction before utilities halted the project in 2017. The rest never produced a single operating reactor.

The failure wasn't about the technology itself. Nuclear projects died in the delivery, the years and billions between an application and a working plant. When Vogtle's second new unit finally entered service in 2024, no other reactor was under construction anywhere in the country.

Can NuScale Solve What Killed the Last Boom?

NuScale's pitch is aimed directly at that problem. The company's 77-megawatt reactor modules are built in a factory rather than assembled on site, and they can be deployed in configurations of up to 12 modules per plant. A 12-module plant would produce 924 megawatts, less than a single new Vogtle unit but potentially faster and cheaper to build.

The company holds the only small modular reactor (SMR) design certification the NRC has issued, and it received approval for an updated design in May 2025. NuScale says it has built a supply chain of more than 60 partners and has executed over 30 supply agreements. CEO John Hopkins framed the challenge plainly: "The question for off-takers is no longer whether to go with nuclear, it is which technology can actually deliver, and when".

But the financials tell a different story. Second-quarter 2026 revenue came in at just $75,000, down from $8.1 million a year earlier, when NuScale was still collecting engineering fees from its Romanian project work. That work wrapped up in late 2025. For the first half of 2026, the company's total revenue was only $640,000. The company's second-quarter net loss attributable to its Class A shareholders was $47.5 million.

How Is NuScale Staying Afloat Without Revenue?

NuScale does hold $1.9 billion in cash and investments, which gives the company runway to fund itself for years. But that cushion has come entirely from shareholders. The weighted-average Class A share count nearly tripled year over year to about 365 million shares, and the company added a new $750 million at-the-market stock sale program in August 2026.

The market is essentially paying for what NuScale might build, not for anything it sells today. That's not automatically a mistake, but the last nuclear cycle offers a cautionary tale. The test for a growth stock like NuScale is clear: not interest, not agreements to study, but a signed, funded order.

• Tennessee Valley Authority Deal: NuScale is in discussions with ENTRA1 Energy, its commercialization partner, toward a definitive power purchase agreement that the company says would potentially be the largest nuclear deployment program in U.S. history, but the deal remains unsigned.
• Romania's RoPower Project: The six-module RoPower project, described as the most advanced SMR effort in Europe by NuScale, is still working through conditions attached to a shareholder vote to advance it.
• New Class of Buyer: Unlike the last boom, this cycle has data-center operators with urgent power needs and deep pockets, but neither of NuScale's two lead opportunities has reached a signed, funded order yet.

Why the AI Data Center Difference Might Matter

The last nuclear boom had committed utilities, federal support, and 26 proposed reactors on file. It ultimately produced two reactors, both late and far over budget. This cycle does have something new: a class of buyer in AI data-center operators with urgent power needs and deep pockets. Artificial intelligence (AI) systems require enormous amounts of electricity to run, and companies like those powering large language models (LLMs) are desperate for reliable, carbon-free power sources.

Meanwhile, other advanced reactor developers are moving faster. Aalo Atomics achieved criticality on July 4, 2026, reaching a sustained chain reaction in just eight months, one of the fastest reactor builds in modern American history. The company built, licensed, and operated its test reactor facility from scratch at Idaho National Laboratory, demonstrating a full-scale core intended for a 10-megawatt reactor.

"We are proud to play a major role in America's nuclear renaissance, going from breaking ground to a sustained chain reaction in just eight months, one of the fastest reactor builds in modern American history," said Yasir Arafat, president and chief technical officer of Aalo Atomics.

Yasir Arafat, President and Chief Technical Officer at Aalo Atomics

Aalo's ultimate objective is to include multiple full-scale power reactors in a 50-megawatt commercial "Aalo Pod" to power AI data centers. The company is already expanding into a one-million square-foot factory to apply assembly-line manufacturing to reactor production, which it says will open the door to mass-producing the Aalo Pod.

The Valuation Question: Is $4.2 Billion Justified?

A $4.2 billion valuation on $10.7 million of trailing sales is arguably priced for the moment the orders arrive. But in the last cycle, getting the order turned out to be the easy part. The hard part was actually building the reactor on time and on budget. NuScale's success depends not just on signing deals with data-center operators, but on proving it can deliver what utilities and tech companies cannot: a reactor that costs what it's supposed to cost and arrives when promised.

For investors and the broader energy industry, NuScale's trajectory will be a crucial test. If the company can convert its two pending deals into signed, funded orders and then actually build reactors faster and cheaper than the Vogtle experience, it could reshape nuclear energy's role in powering AI. If it cannot, the $4.2 billion valuation will look like another chapter in a long history of nuclear promises that failed in the delivery.